| The assessment and qualification of reliability is an important work in the development, design, finalization, purchase, application of weapon system. With the constant improvement of technique level and complexity equipment, the cost and test charge of complex equipment is expensive, the test of which is successive but the test number is not enough and the test conditions are different. Therefore, for the reliability test and assessment of equipment which has the characteristics of small sample, multi-stage, and dynamic population, it's difficult to give scientific and rational result in traditional statistical analysis method.Aimed at the engineering application requirement and theory requirement for reliability test and analysis in the equipment development, this dissertation carries on a systemic research on a series of difficult problem from a completely new perspective, including Bayes fuse information, reliability-growth plan, optimal selection of test scheme for a Bayes plan and reliability assessment of dynamic distribution parameter. A set of theories and methods of Bayes analysis of dynamic distribution parameter is established by incorporating change statistics theory and Bayes method, which provides technical support for equipment development. The main contributions of this dissertation are summarized as follows:1. Firstly, considering the demand of equipment development and reliability specification, this paper analyze the characteristics of development test and its problems in engineering, the above research shows that the representation of prior distribution, modeling of reliability and fuse information are the key problems in Bayes analysis of dynamic distribution parameter. Secondly, based on the Bayes analysis of dynamic distribution parameter,the reliability test and assessment flow chart and its selection principle is constructed. Thirdly, expert's information fusion model based on the D-S principium and optimization model is put forward and the Bayes reliability information fusion approach such as order relation model and conversion of reliability test information are presented. Finally, in this paper the dynamic distribution parameter modeling methods such as order relation model and stochastic process are studied.2. Aimed at the characteristics of equipment development by stages and by batch,Bayes reliability-growth programming model which combines test revised strategy is proposed based on exponential distribution and the non-homogeneous Poisson process. In the case of different stages and different reliability-growth levels, the exact testing information is obtained, and dynamic reliability-growth program and reliability-growth management is implemented.3. A Bayesian reliability growth models of diverse populations based on the new Dirichlet prior distribution is studied. Aiming at some history and expert information during the equipment development, Bayes reliability growth model is presented based on the new Dirichlet distribution, the model can be used to predict the product reliability, which extends application range of the model. The method for determining prior distribution parameters is given by optimization method, it solves the problem of how to verify the hyper-parameters of the new Dirichlet prior distribution as these parameters have no specific physical meaning. Furthermore, it also establishes Bayesian model that can be applied to the reliability growth of Exponential distribution and Weibull distribution products. Then, Bayes point assessment and confidence lower limit on product reliability at current stage are imputed by comprehensively making use of the MCMC algorithm. The results show that the model can not only estimate the reliability growth, but also predict the reliability of posterior development period, which overcomes the defect that the general Bayes model can only estimate reliability growth.4. A bayes plan of reliability qualification test is discussed in combination with the results obtained from the actual tests. On the premise that the quantity of the test is ensured, a scheme of reliability acceptance test in binomial case is formulated by using prior information efficiently. In view of the meaning of sampling risk actually concerned by users and producers in acceptance test, the difference and relationship between the Bayes reliability qualification and assessment are analyzed.The outcome of the project instance is compared with the traditional method according to the plan of reliability acceptance test, the range of Bayes method is clear and definite. Considering not only the testing information from related products but also differences between such products, a new mixed prior distribution is used by introducing the inheritance factor, moreover the inheritance factor is thought as a random variable, then the Bayes decision model of the qualification plan was established. Based on the method above, the Bayes plans of qualification test are researched for binomial case and exponential case. The design experiments show such algorithm is very effective and efficient, and it has active guidance meaning and applied value to the engineering practice.5. The methods of reliability synthetical test design for Bayes analysis of dynamic distribution parameter are applied to reliability test and analysis of the weapon system, which has the characteristics of small sample, multi-stage, dynamic population.In summary, with the aid of related key projects of armament, this dissertation investigates the problem of Bayes reliability analysis of dynamic distribution parameter systematically and deeply in theory, method and application. The studying results and improvements based on this paper will provide a set of perfect-theory and applicable-engineering theory and methods for Bayes reliability analysis of dynamic distribution parameter. And the results are of important theory value and engineering instruction significance for researches on the equipment reliability support technique, which has the characteristics of small sample, multi-stage, and dynamic population.
|
PDF Full Text Request